mirror of
https://github.com/jambonz/jambonz-feature-server.git
synced 2025-12-20 16:50:39 +00:00
add support for vad to gather and transcribe (#67)
This commit is contained in:
@@ -22,6 +22,10 @@ class TaskTranscribe extends Task {
|
||||
this.interim = !!recognizer.interim;
|
||||
this.separateRecognitionPerChannel = recognizer.separateRecognitionPerChannel;
|
||||
|
||||
/* vad: if provided, we dont connect to recognizer until voice activity is detected */
|
||||
const {enable, voiceMs = 0, mode = -1} = recognizer.vad || {};
|
||||
this.vad = {enable, voiceMs, mode};
|
||||
|
||||
/* google-specific options */
|
||||
this.hints = recognizer.hints || [];
|
||||
this.profanityFilter = recognizer.profanityFilter;
|
||||
@@ -105,6 +109,12 @@ class TaskTranscribe extends Task {
|
||||
async _startTranscribing(cs, ep) {
|
||||
const opts = {};
|
||||
|
||||
if (this.vad.enable) {
|
||||
opts.START_RECOGNIZING_ON_VAD = 1;
|
||||
if (this.vad.voiceMs) opts.RECOGNIZER_VAD_VOICE_MS = this.vad.voiceMs;
|
||||
if (this.vad.mode >= 0 && this.vad.mode <= 3) opts.RECOGNIZER_VAD_MODE = this.vad.mode;
|
||||
}
|
||||
|
||||
ep.addCustomEventListener(GoogleTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
|
||||
ep.addCustomEventListener(GoogleTranscriptionEvents.NoAudioDetected, this._onNoAudio.bind(this, cs, ep));
|
||||
ep.addCustomEventListener(GoogleTranscriptionEvents.MaxDurationExceeded,
|
||||
|
||||
Reference in New Issue
Block a user